Mistral AI has launched Le Chonk, its 1.05 trillion-parameter multimodal model, with open weights set to release on October 31, 2026. The model is now available in API preview, marking a significant milestone for European AI development.
TL;DR
- Mistral AI introduces Le Chonk, a 1.05 trillion-parameter model, with open weights to be released on October 31, 2026.
- The model is positioned for programming, cybersecurity, and corporate workflows, offering cost advantages over competitors.
- Le Chonk's open weights could attract enterprises prioritizing data sovereignty, challenging US-based models like GPT-5.5 and Claude Opus 5.
What happened
Mistral AI unveiled Mistral Large 4, nicknamed Le Chonk, a multimodal model built on a mixture-of-experts architecture with 1.05 trillion total parameters and 49 billion active per token. The model is live in public preview through Mistral's API, priced at $1.36 per million input tokens and $4.18 per million output tokens. The full open weights release is scheduled for October 31, 2026.
The nickname 'Le Chonk' originated from a joke on Reddit and X, where fans invented a fictional 30-trillion-parameter model. Mistral CEO Arthur Mensch embraced the meme, and the name stuck. The model's specifications include a 1.6-billion-parameter vision encoder, a one-million-token context window, and fluency in over 160 languages.
Why it matters
Le Chonk's open weights could attract enterprises prioritizing data sovereignty, challenging US-based models like GPT-5.5 and Claude Opus 5. This move could significantly impact the competitive landscape, especially for European enterprises concerned about data control.
While Le Chonk trails behind frontier models like Claude Opus 5 and GPT-5.5 in hard reasoning and agentic benchmarks, it offers cost advantages. The model is priced significantly lower than its competitors, making it an attractive option for enterprises running thousands of API calls daily. Mistral claims Le Chonk leads on AutomationBench, a workflow-automation benchmark, at around a fifth of the per-task cost of rival models.
Key facts
- Mistral Large 4, nicknamed Le Chonk, has 1.05 trillion total parameters and 49 billion active per token.
- The model is available in API preview on October 6, 2026, with open weights to be released on October 31, 2026.
- Le Chonk is priced at $1.36 per million input tokens and $4.18 per million output tokens.
- The model includes a 1.6-billion-parameter vision encoder and a one-million-token context window.
- Mistral claims fluency in over 160 languages for Le Chonk.
- The model was trained on a cluster of 3,800 Nvidia Grace Blackwell GPUs.
- Le Chonk scored 38 on Artificial Analysis's index, trailing DeepSeek's 4.1 Flash.
- The model is positioned for programming, cybersecurity, and corporate workflows.
Context
Mistral AI's release of Le Chonk with open weights is a significant development in the AI landscape. It challenges the dominance of US-based models like GPT-5.5 and Claude Opus 5, offering a European alternative that prioritizes data sovereignty. This move could attract enterprises concerned about data control and looking for cost-effective solutions.
The release of Le Chonk also highlights the growing capabilities of European AI labs. Mistral's ability to train and release a trillion-parameter model demonstrates significant advancements in AI technology outside the US. This could spur further innovation and competition in the global AI market.
